statement of qualifications-cfd

8/7/2019 Statement of Qualifications-CFD

1/77

. . . . . . . .

........

allied environmental technologies, inc.

COMPUTATIONAL

FLUID

DYNAMICS

MODELING

APPLICATIONS

FOR

ENGINEERING SOLUTIONS

STATEMENT OF QUALIFICATIONS

15182 Bolsa Chica Road, Suite D

Huntington Beach, CA 92649 USA


2/77

i Statement of Qualifications

CFD for Engineering Solutions ABSTRACT

Computational Fluid Dynamics

forEngineering Solutions

ABSTRACT

Computational Fluid Dynamics (CFD) is becoming a critical part of

the design process for more and more companies. CFD makes it possible

to evaluate velocity, pressure, temperature, and species concentration of

fluid flow throughout a solution domain, allowing the design to be opti-

mized prior to the prototype phase.

At Allied Environmental Technologies, Inc. our philosophy embrac-

es the idea that advanced Computational Fluid Dynamics (CFD) technolo-

gy combined with an intelligent graphical interface results in a powerful

simulation capability. Allied offers solutions to industry in aerodynamics,

chemical, and combustion processes, environmental flows, and more. At

the core of the CFD modeling is a three-dimensional flow solver that is

powerful, efficient, and easily extended to custom engineering applica-

tions. We combined the Fluents proven simulation capabilities in fluid

flow, heat transfer, and combustion with years of practical experience in

power and other industries.


3/77

ii Statement of Qualifications

CFD for Engineering Solutions ABSTRACTIn designing a new mixing device, injection grid or just a simple gas

diverter or a distribution device, design engineers need to ensure ade-

quate geometry, pressure loss, and residence time would be available.

More importantly, to run the plant efficiently and economically, operators

and plant engineers need to know and be able to set the optimum para-

meters. For example, to run the injection grid for cooling purposes, opera-

tor needs to know and be able to set the liquid flow through the orifices to

achieve optimum droplet size to ensure a complete evaporation and prop-

er mixing parameters. On the other hand, while running the same grid

with the gaseous media injected, the operator needs to be concerned with

the time required for achieve the proper mixing parameters only. For

these cases, CFD tools are very useful.

In a multiphase flow model, equations for conservation of mass,

momentum, and energy are solved for each phase (gas bubbles, solid

particles and liquid). Additional equations are included to describe the in-

teractions between the phases due to drag, turbulence and other forces.

From the computed solutions, information on velocity, temperature, and

concentration of each phase is available for the whole vessel. Other pa-

rameters, such as mixing time, gas hold-up and particle concentration, can

be deduced from the computer solutions.


4/77

iii Statement of Qualifications

CFD for Engineering Solutions EXECUTIVE SUMMARY

TABLE OF CONTENTS

ABSTRACT .................................................................................................................. i

TABLE OF CONTENTS ............................................................................................. iii

EXECUTIVE SUMMARY ............................................................................................ v

FLOW ENGINEERING................................................................................................V

CHAPTER 1.BACKGROUND ................................................................................1-1

1.1 COMPUTATIONAL FLUID DYNAMICS MODELING ............................................ 1-1

1.1.1 Fluid Mechanics. ............................................................................... 1-2

1.1.1.1 Laminar and Turbulent Flows. ................................................... 1-2

1.1.1.2 General Theory and Principal Equations. ................................. 1-3

1.1.1.3 Boundary Conditions ................................................................. 1-6

CHAPTER 2.REFERENCES ..................................................................................2-1

CHAPTER 3.CASE STUDIES ................................................................................3-1

3.1 FILTER HOUSE. ..........................................................................................3-2

3.1.1 Model Setup. .....................................................................................3-2

3.1.2 Results Discussion. ........................................................................... 3-3

3.1.3 Figures. .............................................................................................3-3

3.2 COMBUSTION GAS TURBINE EXHAUST SYSTEM ........................................... 3-1

3.2.1 Geometry and Boundary Conditions ................................................. 3-1

3.2.2 Results and Discussion for Uniform Inlet .......................................... 3-3

3.2.3 Results and Discussion for Non-Uniform Inlet .................................. 3-4

3.2.4 Conclusions .......................................................................................3-6

3.3 HOT GAS PRECIPITATOR ............................................................................ 3-8

3.3.1 Gas Flow in Electrostatic Precipitators .............................................. 3-8

3.3.1.1 IGCI EP-7 Standard .................................................................. 3-9

3.3.2 Electrostatic Precipitators ................................................................ 3-10

3.3.3 Model Setup. ................................................................................... 3-13

3.3.4 Results Discussion. ......................................................................... 3-13


5/77

iv Statement of Qualifications

CFD for Engineering Solutions EXECUTIVE SUMMARY3.3.5 Summary and Conclusions ............................................................. 3-14

3.4 AIR FLOW MODELING IN THE REGENERATIVE THERMAL OXIDIZER .............. 3-23

3.4.1 Packed Tower .................................................................................. 3-23

3.4.2 Results ............................................................................................3-23

3.4.3 Summary and Conclusions ............................................................. 3-24

3.4.4 Figures ............................................................................................3-24

3.5 MODEL STUDY FOR IN-DUCT BURNERS ..................................................... 3-26

3.5.1 Model Set-up ...................................................................................3-26

3.5.2 Discussion .......................................................................................3-26

3.5.3 Figures ............................................................................................3-27

3.6 LINDBERG INCINERATOR AND ABSORPTION TOWER FLOW EVALUATION...... 3-32

3.6.1 Model Setup ....................................................................................3-32

3.6.2 Results and Discussion ................................................................... 3-34

3.6.3 Conclusions .....................................................................................3-35

3.6.4 Figures ............................................................................................3-35


6/77

v Statement of Qualifications

CFD for Engineering Solutions EXECUTIVE SUMMARY

EXECUTIVE SUMMARY

With the appearance of powerful and fast computers, new possibili-

ties for replacing time-consuming model testing and field-testing have ari-

sen. This involves solving the differential equations describing fluid mo-

tion, using either a finite volume or sometimes, but more rarely due to

larger amount of CPU time required, a finite element method. These me-

thods applied for the solution of the fluid equations of motion are named

computation fluid dynamics or simply CFD.

CFD essentially takes the momentum, heat and mass balance eq-

uations, along with other models describing the equipment performance,

and solves them to give information such as temperature profiles, velocityprofiles and equipments size. The basic principle of the CFD modeling

method is that the simulated flow region is divided into small cells within

each of which the flow either kept under constant conditions or varies

smoothly. Differential equations of momentum, energy, and mass balance

are discretized and represented in terms of the variables at the center of

or at any predetermined position within the cells. These equations are

solved iteratively until the solution reaches the desired accuracy.

FLOW ENGINEERING

Allied Environmental Technologies, Inc. Flow Engineering offers a

variety of services, which specifically support CFD model studies. We can

conduct internal inspections to document areas of particulate buildup,


7/77

vi Statement of Qualifications

CFD for Engineering Solutions EXECUTIVE SUMMARYwear patterns, and existing internal structural members, and flow devices.

Filed airflow measurements within precipitators, scrubbers, SCR vessels,

and flues could be conducted to evaluate flow problems prior to conduct-

ing a model study, or at the completion of a rebuild project to satisfy oper-

ational guarantees.

In addition to direct measurements and the CFD analyses, various

methods of flow visualization could be employed. This visualization soft-

ware helps to quickly understand the most complex features of ones simu-

lation and effectively communicate ones conclusions.


8/77

1-1 Statement of Qualifications

CFD for Engineering Solutions CHAPTER 1. BACKGROUND

CHAPTER 1.BACKGROUND

To gain acceptance and confidence in applying the CFD technolo-

gy, numerous validation exercises were conducted in which CFD solutions

were compared against measured values from experiments.

1.1COMPUTATIONAL FLUID DYNAMICS MODELING

Traditional restrictions in flow analysis and design limit the accuracy

in solving and visualization fluid-flow problems. This applies to both sin-

gle- and multi-phase flows, and is particularly true of problems that are

three dimensional in nature and involve turbulence, chemical reactions,

and/or heat and mass transfer. All these can be considered together in

the application of Computational Fluid Dynamics, a powerful technique

that can help to overcome many of the restrictions influencing traditional

analysis.

The basic principle of the CFD modeling method is simple. The

flow regime is divided into small cells within each of which the flow either

kept constant or varies smoothly (the smooth variation is normally as-

sumed for modern techniques that normally have order of accuracy above

one). The differential equations of momentum, energy, and mass balance

are discretized and represented in terms of the variables at the cell cen-

ters (collocated variables arrangement) or at a predetermined position

within the cells (normally cell walls for velocities staggered variable ar-

rangement). These equations are solved until the solution reaches the

desired accuracy.


9/77


CFD for Engineering Solutions CHAPTER 1. BACKGROUNDCFD modeling provides a good description of flow field variables,

velocities, temperatures, or a mass concentration anywhere in the region

with details not usually available through physical modeling. It is especial-

ly useful for determining the parametric effects of a certain process varia-

ble. Once the basic model is established, parametric runs can usually be

accomplished with reduced effort. In addition, CFD can be used to simu-

late some of the hard to duplicate experimental conditions or to investigate

some of the hard to measure variables.

The compartment volume under study is divided into an array of

small cells, and the equations of momentum, energy, and mass transfer

are solved numerically within each cell. This provides a good description

of flow field variables such as velocities, temperatures, and mass concen-

trations everywhere in the region. The boundary conditions for gas flow

could be specified according to the thermodynamic, heat, and mass trans-

fer characteristics at different gas flows and velocities. The mixing, distri-

bution, and stratification could be then verified by inspecting the calculated

pressure distribution across the exhaust duct.

The computational results, including the three-dimensional distribu-tion of the flue gas velocity (flow) and the total pressure loss, are pre-

sented in both mathematical and graphical interpretation.

1.1.1 Fluid Mechanics.

1.1.1.1 Lam inar and Turbulent Flows.In fluid mechanics, it is customary and useful to classify fluid flow

into several major categories, which for subsonic conditions usually are

taken as ideal flow, laminar flow, and turbulent flow. Ideal flow refers to

the flow of so-called ideal or frictionless fluids and is the type developed

theoretically to a high state of mathematical perfection under the general

designation of classical hydrodynamics. Although much criticized as a

perfectionist and useless art, the results of this highly theoretical develop-


10/77


CFD for Engineering Solutions CHAPTER 1. BACKGROUNDment have surprisingly wide range of practical applications in such broad

fields as aerodynamics, wave motion, and supersonics. For example,

even functioning of such commonplace devices as the Pitot tube depends

on the theoretical results of classical hydrodynamics.

1.1.1.2 General Theory and Principal Equa tions.Observations show that in any system involving fluid flow (either

gas or liquid) two completely different types of this flow can exist. Rey-

nolds first observed it in 1883 in his experiment in which water was dis-

charged from a tank through a glass tube. A valve at the outlet could con-

trol the rate of flow, and a fine fragment of die injected at the entrance of

the tube. At low velocities, it was found that the dye fragment remainedintact throughout the length of the tube, showing that particles of water

moved in parallel lines. This type of flow is known as laminar, viscous or

streamline, the particles of fluid moving in an orderly manner and retaining

the same relative positions in successive cross-sections.

As the velocity of the fluid was increased by the opening of the out-

let valve, a point was eventually reached at which the dye fragment at first

began to oscillate and then broke up. That resulted in a fact that the colorwas diffused over the whole cross-section. Thus, showing that the par-

ticles of fluid no longer moved in an orderly manner but occupied different

relative positions in successive cross-sections. This type of flow is known

as turbulent and is characterized by continuous small fluctuations in the

magnitude and direction of the velocity of the particles, which are accom-

panied, by corresponding small fluctuations of pressure.

When the motion of a fluid particle an a stream is disturbed, its iner-

tia will tend to carry it on the new direction, but the viscous forces due to

surrounding fluid will tend to make it conform to the motion of the rest of

the stream. In viscous flow, the viscous shear stresses are sufficient to

eliminate the effects of any deviation, but in turbulent flow, they are inade-


11/77


CFD for Engineering Solutions CHAPTER 1. BACKGROUNDquate. The criterion, which determines whether flow will be viscous or tur-

bulent, is therefore the ratio of the internal force to the viscous force acting

on a particle.

By dividing the characteristic inertia forces of the fluid flow to cha-

racteristic viscous forces of this flow, it is possible to find out that the ratio

is proportional to dimensionless value

where:

is fluid density,

v is characteristic fluid velocity,

is characteristic length of the fluid flow and

is dynamic viscosity of the fluid.

This dimensionless value Re is known as Reynolds number. The

value of the Reynolds number at which the flow ceases to be laminar and

becomes turbulent depends on the flow geometry. However, usually the

transition from laminar to turbulent flow occurs in the range 1,000-10,000

of the values of Reynolds number Re. Thus, to solve the problem accu-

rately, we must provide the appropriate model that is able of handling the

turbulent flows.

To obtain the stationary fields of velocity, pressure, and/or tempera-

ture for any fluid flow one has to solve the equations of motion of the fluid

together with energy equation and equation of state of the fluid under ap-

propriate boundary conditions. Theoretically, the equations of motion for

both laminar and turbulent Newtonian flows are the same Navier-Stokes

equations. However, in practice the solution of Navier-Stokes equations

for the turbulent flow requires very small computational cell size due to

necessity to accurately simulate small-scale turbulent eddies. This leads

to very large computer memory consumption, which is possible only on

modern-day supercomputers. Moreover, this kind of precision normally is

not required for the engineering system where usually only time-averaged


12/77


CFD for Engineering Solutions CHAPTER 1. BACKGROUNDvalues of the flow are important and small turbulent velocity and pressure

calculations are immaterial.

To avoid the unnecessary consumption of the computer time for the

solution of the full-scale Navier-Stokes equations for the turbulent flow

number of numerical techniques have been proposed. These are directed

to determining the time-averaged values of the velocities and pressure in

turbulent flows rather than instantaneous values of these parameters. The

most popular for engineering problems among these techniques are so-

called effective or turbulent viscosity techniques. The essence of effective

viscosity techniques is that the viscosity used in the equations of motion of

the fluid in the turbulent regime is not the real fluid viscosity (laminar vis-

cosity), which does not depend on the parameters of the flow, but some

effective viscosity which is determined according to the parameters of the

turbulent flow and thus is no material constant and which is usually larger

than the laminar viscosity. The introduction of this effective viscosity al-

lows us to determine the field of the mean turbulent parameters of the flow

in similar way as it is done for the parameters of the laminar flow. While

the effective viscosity techniques are not perfect and sometimes produce

wrong solutions, they have been around for several decades already and

all their shortcomings are well documented in the literature. At the present

day, the effective viscosity techniques are perhaps the only techniques

used for the solution of the engineering problems, which involve flow of

the fluids in the turbulent regimes.

One of the most popular effective viscosity models for the simula-

tions of turbulent flow historically was the standard high Reynolds number

Harlow-Nakayama k- model of the turbulent flow incorporated into the

CFD software package has been used. The model provides the time-

averaged values of velocities and pressure of the air throughout the sys-

tem. This model has become perhaps the most popular two-equation

model of turbulence, though mostly for historical rather than scientific rea-


13/77


CFD for Engineering Solutions CHAPTER 1. BACKGROUNDsons. K- model calculates the effective viscosity associated with the tur-

bulence based on the solution of two equations for the density of turbulent

kinetic energy k and rate of the dissipation of this energy . Since two ad-

ditional differential equation have to be solved in this model, it belongs to

the class of two-equation turbulence models. Like other two-equation

models, the k- model is free from the need to prescribe the length scale

distribution.

1.1.1.3 Bounda ry Condit ionsIn many reaction systems, determining the boundary conditions is

more difficult than performing the time-dependent simulation. It is not very

meaningful to calculate a time-dependent event in a system if one doesnot know how the initial state ha occurred, even if the initial state is known.

Calculating a time-dependent event always must start from a stationary

state.

The high-Reynolds number k- model becomes invalid near the

walls and thus has to be regarded with the set of wall functions (low-

Reynolds number models are also available but rarely used in engineering

community). Allied Environmental Technologies, Inc. uses the approachof the logarithmic wall functions based on the ideas of Schlichting. These

functions are applicable for the Reynolds numbers based on the length of

the system, which are in the range from approximately 106

to 109. The

structure of these functions depends strongly on the roughness of the

walls since the increased roughness actually promotes development of the

turbulence in the boundary layer and thus changes flow characteristics. In

addition, even in the places where turbulent flow is fully developed thecharacteristic size of the wall roughness could be bigger that the thickness

of this boundary layer. In this case, the flow will be greatly influenced by

the wall roughness. Unfortunately, for the most cases the actual value of

the wall roughness is not known. In cases like that, where the actual val-

ue of the wall roughness is not known it has been standard CFD practice


14/77


CFD for Engineering Solutions CHAPTER 1. BACKGROUNDto ignore the wall roughness at all and to assume the walls of the system

to be smooth ones. In our simulation, we followed this standard practice

for unknown roughness wall boundary conditions.

The other boundary conditions applied are fixed mass flow rate at

the exit of the system and fixed static pressure at the entrance to the sys-

tem. The inlet boundary conditions require the prescription of the two tur-

bulent parameters, density of the kinetic energy of turbulence k and rate of

the dissipation of this energy . These values are unknown, however they

can be determined if we assume the intensity of the turbulence at the en-

trance. This allows us to determine the value of k, the value of is than

determined based on the inlet linear dimensions. The inlet turbulence in-

tensity is normally assumed to be between 1% and 5%, 2% value has

been chosen for our calculations.

The equation of state of air was assumed the one of the constant

density gas. This assumption is valid since the static pressure drop

through the system is of order of several inches of water. This value is

less than 1% of the atmospheric pressure, and the expected density

change is less than one percent. Braden Manufacturing has provided theflow rate in the Filter House. The entrance conditions of the air were as-

sumed the standard ones.


15/77


CFD for Engineering Solutions CHAPTER 2. REFERENCES

CHAPTER 2.REFERENCES

Following is an abbreviated list of our customers:

1. Braden Manufacturing, LLC

2. GE Power Systems

3. Lurgi GmbH and Lurgi, Inc.

4. TurboSonic, Inc.

5. AES, Inc.

6. Savage Zinc, Inc.

7. Deltak

8. GEA, Inc.

9. Nanticoke Generating Station, Ontario Power

10. Mirant, Inc.


16/77


CFD for Engineering Solutions CHAPTER 3. CASE STUDIES

CHAPTER 3.CASE STUDIES


17/77


CFD for Engineering Solutions CHAPTER 3. CASE STUDIES3.1FILTER HOUSE.

3.1.1 Model Setup.

The Filter House configuration includes four filter units placed in ar-

row formations with each unit containing a pre-filter, a filter, and a mist

eliminator. In the present simulation, as in the previous one, the special

care has been taken to include fine structural details in the simulation.

These details include filter housing hood, details of the filter and pre-filter

grid, cooler tanks, support columns, half-pipe situated at the top of the fil-

ter housing exit and vertical-horizontal braces according to the drawings

provided by the customer.

The computational mesh has been constructed using the specifica-

tions provided by the customer using the body-fitted coordinates ap-

proach, which allows the creation of the grid of complex geometry. The

computational grid consisted of 52 32 x 28 or 46,592 computational

cells. The Figures below depict the structure of the computational cells for

the simulations considered. In fact, only one half of the Filter House duct-

work has been actually simulated, which was made possible by a com-

plete symmetry with respect to the vertical plane passing through the cen-

ter of the Filter House.

Computational Grid. View 1. Computational Grid. View 2.


18/77



3.1.2 Results Discussion.

A number of cases had been evaluated. In the present simulation,

complicated patterns of perforated screens with variable permeability were

installed in the Plane A (downstream of evaporative cooler modules). The

whole length of the cooler modules was divided into four parts (Zones);

each Zone being 8.5 feet long.

Zone 1 (closest to the exit from the module) contains screens with 20%, 25% and45% open space;

Zone 2 contains screens with 20%, 30% and 45% openspace;

Zone 3 is covered by the screens with 45% open space, and Zone 4 - the last part, has no screens at all.

The computational results, including the three-dimensional distribu-

tion of the flue gas velocity (flow) are presented graphically below. The

maximum velocity in the Plane A was computed to be 832 ft/min, while the

total pressure loss across the filter house is 2.19 inches of water.

Furthermore, the results of the gas velocity calculated for all cases

in the selected plane C are presented in a form of Contour Lines. There

are six contour lines (where appropriate values are reached) colored from

blue to red. These lines correspond to values:

#1 500 ft/min

#2 600 ft/min

#3 700 ft/min

#4 800 ft/min

#5 900 ft/min

#6 1000 ft/min.

3.1.3 Figures.


19/77



Plane A velocity values (ft/min). Plane B velocity values (ft/min)

Plane C velocity values (ft/min). Plane D velocity values (ft/min).

Plane E velocity values (ft/min). Plane F velocity values (ft/min).

Plane G velocity values (ft/min). Plane H velocity values (ft/min).


20/77



Velocity distribution in the horizontal plane situated at 1/12height of the filter housing from the bottom of the modules.

Velocity distribution in the horizontal plane situated at 1/4height of the filter housing from the bottom of the mod-

ules.

Velocity distribution in the horizontal plane situated at 5/12

height of the filter housing from the bottom of the modules

Velocity distribution in the horizontal plane situated at 7/12

height of the filter housing from the bottom of the mod-ules.

Velocity distribution in the horizontal plane situated at 9/12height of the filter housing from the bottom of the modules.

Velocity distribution in the horizontal plane situated at11/12 height of the filter housing from the bottom of the

modules


21/77



Case 1. Plane C Velocity contour lines (ft/min). Case 2. Plane C Velocity contour lines (ft/min).



Case 7. Plane C Velocity contour lines (ft/min).


22/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONS3.2COMBUSTION GAS TURBINE EXHAUST SYSTEM

3.2.1 Geometry and Boundary Conditions

A computational domain for the gas turbine exhaust system is shown

in Figure 1 below at different viewing angles. The blue region is the inlet

where the exhaust gas from the gas turbine enters the diffuser duct. The

diameter at the inlet is 4.877 m (16). The inlet is then connected with a

diffuser duct that gradually transforms the circular duct to a rectangular

duct of larger cross-sectional area. In the rectangular section, nine iden-

tical silencer plates are visible on the plot. The panels have a length of

6.045 meters (19 10). At the upstream direction, there is a half cylinder

of 0.254-meter (10) radius. The width of all the panels is 0.508 m (20).The gap between the panels is 0.305 m (12). The gap between the out-

most panel and the external wall is 0.152 m (6).

Figure 1. Cut-off view of the computational domain. The blue region is the inlet where the exhaust gas from thegas turbine enters the diffuser duct. The silencer plates are visible. At top of the stack, the outlet is represented

with a red region

The entire computational domain was divided into a number of struc-

tured cells. The dimension of the mesh is 43x72x56. The total number of

cells is 173,376.


23/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSThe exhaust gas has a temperature of 890.5K (1143.5F). The pres-

sure is 99632.75 N/m2 (14.45 psia). The molecular weight of the gas is

28.10, and the density is 0.378 kg/m3. The viscosity of the gas is set to be

39.227x10-6 N sec/m2. The incoming gas flow rate is 892.9 Lb/sec

(405.86 kg/sec), which corresponds to an average incoming velocity of

6.1046 m/sec at inlet. The exhaust gas from the last stage GT bucket has

a swirl angle of 4.22 degrees.

Two cases were considered for the inlet velocity profile. For the

uniform inlet case, a constant incoming u (x-direction) velocity component

was specified. While the v (y-direction) and w (z-direction) according to

the swirling angle. Whereas, for the non-uniform inlet case, u velocity val-ues at the following four locations are specified:

R/R0=0.00 R/R0=0.38 R/R0=0.70 R/R0=1.00U/Vave 0.55 1.09 1.37 0.10

The Lagrange Interpolation formula was used to determine the u

velocity at other radial positions. After the u velocity is calculated, the v

and w velocity is determined according to the swirling angle.

Though the ambient temperature is 95o

F, the computational domain iswrapped with insulation and the calculation is done with adiabatic no-slip

boundary condition for all internal surfaces. Furthermore, the walls are

assumed to be smooth. The pressure variation within the domain is rela-

tively small compared to the ambient pressure. Thus, the fluid is approx-

imated with incompressible fluid with the properties as specified above.

The k-e two-equation turbulence model is applied to simulate the

turbulent flow. The incoming gas was assumed to have 1% free stream

turbulence kinetic energy. The dissipation rate is determined by using the

above turbulence kinetic energy with the inlet diameter (4.877 m) as the

characteristic length scale. Unless known quantities are measured at the

inlet for the turbulence properties, this is a common practice for k-e treat-


24/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSment at inlet and has been applied for many turbulent flow simulations.

The standard wall function treatment is adopted for all non-slip walls.

3.2.2 Results and Discussion for Uniform Inlet

Figure 2 below present three-dimensional cutoff views of the geometry

with two perpendicular cut planes showing the flow patterns.

The friction effects from the no-slip walls and the diverging diffuser

duct geometry lead to non-uniformity of velocity at the edges. However,

the velocity at the center portion is virtually uniform. The effects of the si-

lencer panels can be seem from the plots below. The graph on the left re-

veals that the center channels have higher velocity magnitude. Further-

more, the high velocity region shifted toward higher elevation (+Z direc-

tion). It is because the flow is about to turn upwards in the vertical stack.

It is also because the bottom wall inclines upwards behind the silencer

panels.

Figure 2. The cutoff view of the exterior walls with the silencer panels. There are two cut planes of velocity vec-tors showing the velocity vectors through the center symmetric plan and horizontal velocity vectors. The colorsof the vectors indicate the magnitude of the velocity.

Within each flow channel, the flow shows similar pattern that has

higher center velocity. It is characteristics for flow between parallel plates.


25/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSThe 3-D effects are obvious on the plots. The following Figure 3 depicts the

top view of a cut plane of flow through the centerline of the inlet (Z=0). The in-

coming uniform flow expands as it enters diffuser duct. The existence of the si-

lencer panels reduces the area available to the flow. Thus, the velocity magni-

tude in the center of the channels speeds up. The center channels have higher

averaged velocity than the outside channels. This confirms the same observa-

tion from the previous Figures.

Figure 3. This is the velocity distribution at the end ofthe silencer panels. The upper portion (+Z-direction)has higher velocity than those at lower portion.

Figure 4. This is the top view of a cut plane of flowthrough the centerline of the inlet. The incoming uni-form flow expands as it enters diffuser duct. The exis-tence of the silencer panels reduces the area availableto the flow. Thus, the velocity magnitude in the centerof the channels increases. The center channels havehigher averaged velocity than the outside channels.The jets exiting from the channels are also obvious inthis plot.

3.2.3 Results and Discussion for Non-Uniform Inlet

The Figure 5 presents three-dimensional cutoff views of the geome-

try with two perpendicular cut planes showing the flow patterns.

The Figure 6 is a horizontal plane cuts through the centerline of the in-

let. This is the Z1-Z1 plane similar to the one before. The Figure 7 de-

picts the top view of a cut plane of flow through the centerline of the inlet

(Z=0). The incoming flow distribution is quite visible. After entering the


26/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSdiffuser duct, it expands horizontally (Y-direction) as well as vertically.

The existence of the silencer panels reduces the area available to the

flow. Thus, the center channels have higher averaged velocity than the

outside channels. However, the distribution of flow in all channels is more

even than case with the uniform inlet.

Figure 5. The cutoff view of the exterior walls with the silencer panels. There are two cut planes of velocity vec-tors showing the velocity vectors through the center symmetric plan and horizontal velocity vectors. The colorsof the vectors indicate the magnitude of the velocity. The non-uniformity of inlet velocity can be seen.

Figure 6. This is the velocity distribution at the end ofthe silencer panels. It could be seen that the upperportion (+Z-direction) has higher velocity than those atlower portion.

Figure 7. The Figure above depicts the top view of acut plane of flow through the centerline of the inlet.The incoming non-uniform flow expands as it entersdiffuser duct. The existence of the silencer panelsreduces the area available to the flow. Thus, the ve-locity magnitude in the center of the channels speedsup. The averaged velocity distribution in the channelsseems more uniform the other case that has uniforminlet velocity. The reason is probably due to the rela-


27/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONStively higher inlet flow nears the wall compensates thewall friction. The jets exiting from the channels arealso obvious in this plot.

The next two figures present the vertical plane cut for the uniform

(left) and the non-uniform (right) cases respectively. The non-uniform in-

let flow distribution could be easily seen on the Figure 8.

Figure 8. A vertical cut plane cutting through the centerof the model (Y = 0.0 m). Geometrically, this is thesymmetric plane. However, due to the swirling, this isnot a symmetric plane of the flow. Since this cutsthrough a silencer panel, there is no flow in the silenc-er region. The expansion of the flow at diffuser can beseen on the left of the plot. The flow turn from horizon-tal direction to vertical stack is also shown.

Figure 9. A vertical cut plane cutting through the centerof the model (Y = 0.0 m). Geometrically, this is thesymmetric plane. However, due to the swirling, this isnot a symmetric plane of the flow. Since this cutsthrough a silencer panel, there is no flow in the silenc-er region. The expansion of the flow at diffuser can beseen on the left of the plot. The turning of flow fromhorizontal direction to vertical stack is shown in thisplot.

Overall, the averaged pressure difference between the inlet and

outlet for case 2(non-uniform inlet) is 809 N/m2 or 3.25 inches of water,

while the total pressure loss for this case is 816 N/m2 (or 3.27 inches of

water).

3.2.4 Conclusions

The uniform and non-uniform gas exhaust flows were simulated for the Ex-haust Ductwork System. In both cases, the flow reached fully developed pro-file in the vertical stack. The flow in those regions is uniform in the center ofthe stack with a thin boundary layer. Design of the connecting bend be-tween the horizontal housing and the vertical bend allows the flow to turnfrom horizontal direction to vertical direction without separation. Thoughthere is secondary flow caused by the turning, it reaches fully developed be-fore exit.


28/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONS The silencer panels force the fluid to flow between the parallel walls. How-

ever, the fluid is still allowed to shift upwards as the downstream stack influ-ences it.

The silencers cause the fluid to form jets exiting the channels. The panelsalso induce small recalculation region right behind each panel.

Flow distribution among the channels in the silencer region has similar cha-racteristics for both cases. However, the distribution is more even for thenon-uniform inlet case.

The average pressure drop for both cases is quite similar. For the uniformgas flow at the inlet, the average pressure drop is 3.12 inches of water. Whe-reas, for the Case 2, non-uniform inlet, the average pressure drop is 3.25inches of water.

The total pressure drop is 3.13 and 3.27 inches of water for both

cases respectively.


29/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONS3.3HOT GAS PRECIPITATOR

The entire region of the simulation consisted of the three-

dimensional model of the hot gas electrostatic precipitator. The computa-

tional results, including the three-dimensional distribution of the flue gasvelocity (flow) are presented graphically on the figures below.

The hot gas electrostatic precipitator design considered includes

addition of the variable aperture gas distribution device at the precipitator

inlet. The purpose of this device is to create a controlled non-uniform gas

distribution at the inlet to direct more airflow upwards, thus creating lower

gas velocities at the bottom portion of the precipitator. This device is pre-

sented by a grid of vertical channels, which are situated to allow for larger

gas flow at the top and lower towards the bottom. The geometry also in-

cludes gas distribution grid at the exit from the electrostatic precipitator

and set of baffles and internal walkways.

The inlet grid had been tuned to allow for open areas of 60% and

40% at the top and the bottom respectively with the gradual convergence

in between.

It has been shown that the addition of the variable flow type gas

distribution device at the precipitator inlet does improve the gas flow distri-

bution.

3.3.1 Gas Flow in Electrostatic Precipitators

Gas flow is fundamental to electrostatic precipitation. Its influence

on overall performance is equal to or greater than of the corona and the

electrostatic forces acting on the particles. Type of gas flow determines

the character of particle collection, leading to a deterministic algebraic law

for laminar flow and to a probability exponential law for turbulent flow. Dis-

turbed gas flow, in the form of excessive turbulence, gas jets, swirls, pul-

sations, and other unbalanced erratic flow conditions, in general causes


30/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSheavy reentrainment loss from electrodes and hoppers, as well as poor

collection initially. Evidence from countless case histories shows the

magnitude of these effects on precipitator performance. It is common ex-

perience to increase precipitator efficiency simply by correction and im-

provements in gas flow.

Despite the importance of gas flow and the serious, often disastr-

ous, effects of mal-flow on precipitators operation, scientific investigations

of gas flow phenomena and problems in this field date only from about

1945.

Application of fluid mechanics principles to industrial gas flow band

to electrical precipitation has followed a sporadic, rather than a smooth,

scientific path. Adaptation of results and theories from the relatively so-

phisticated filed of modern aerodynamics has been difficult because of

fundamental differences in conditions and the primitive state of industrial

gas-flow technology1. Actually, the philosophy and methods followed

have been more akin to the semi-empirical hydraulics of the nineteen-

century than to aerodynamics, with the exception of qualitative use of

Prandtls boundary-layer concept and Taylors statistical theories of turbu-lence.

3.3.1.1 IGCI EP-7 StandardThe US International Gas Cleaning institute (IGCI) published EP-7

in November 1970. It deals with standards for gas distribution and mod-

el testing3. In this publication, it is clearly pointed out that the gas distribu-

tion demands are subordinate to the emission guarantee and need not be

corrected, if it does not comply with IGCI EP-7, as long as the emission

guarantee is met.

IGCI EP-7 lays down requirements as to a number of measuring

points, their distribution and the subsequent evaluation and statistics, e.g.

it demands that 85% of the velocities in a cross-section, 3 feet (0.9.m)


31/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSdownstream of the leading edge of the first filed, are less than or equal to

1.15 times the average velocity. Furthermore, 99% must be less than or

equal to 1.40 time average. The same need apply to a cross-section 3

feet (0.9 m) from the outlet of the last field. Specific demands with 1.10

instead of 1.15, and 90% instead of 85% are not motivated by precipitator

physics, but merely in an attempt to demonstrate effective project man-

agement.

In the case of a presumed normal velocity distribution the

[1.15|85%] claim corresponds to a certain coefficient variation,m

, be-

ing the standard deviation and m the mean velocity.

Experience seems to indicate that is almost independent of mean

velocity. In fact, the lower the average velocity, the more difficult it will be

to obtain a good gas distribution, andm

increases with decreasing mean

velocity.

3.3.2 Electrostatic Precipitators

The hot gas precipitators were supplied by Wheelabrator-Frye, Inc.

based on Lurgi design. The precipitators were designed with a single

chamber each having individual inlet and outlet nozzles.

The precipitator has two mechanical and electrical fields in the di-

rection of the gas flow. Each field is 14 feet 1 inches long by the 22 feet

effective collecting plates height. There are 17 gas passages on 12 inch

spacing in each chamber. A common dust-collecting hopper is serving

both fields.

The case study included two gas distribution grids, one at the pre-

cipitator inlet and the other at the precipitator exit. The exit grid consists of

14 vertical channels positioned with the constant width. The inlet grid, on


32/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSthe other hand, although also consisting of 14 vertical channels has those

tapered so that the open area of the grid is gradually converging from 60%

and 40% top to bottom respectively.

At the design operating gas flow of 54,010 acfm, the precipitators

have a specific collecting area (SCA) of 388 ft2/1,000 acfm. Precipitator

design and operating details are described in the Table below.


33/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONS

UNIT ID == > Unit 1

Case ID ===> Design Data Oper./Calc. Data

NO. FGC Option === >

ITEM English Metric English Metric

1 Standard Temperature Deg. F. - Deg. C 32 0 32 02 Conditions Pressure Hg - mm. H2O 29.92 760 29.92 760

3 Temperature Deg. F - C 660 349 660 349

4 Moisture %

5 Site Elevation ft - m 500 152 500 152

6 Design Heat Input MBtu/hr

7 Design Coal Heat Value Btu/lb

8 Inlet Fuel Burn Rate lb/hr

9 Conditions Gas Volume/100 lb coal scf, dry

10 Flue Gas Flow, dry cfm - Nm3/s

11 Flue Gas Flow, wet cfm - Nm3/s

12 Flue Gas Flow, actual acfm - m3/s 54,010 25 54,010 25

13 Fly

14 Ash

15 Concentration gr./acf - g/m3 3.560 8.149 3.560 8.149

16 Loading lb/MBtu

17 General ESP/Blr No. 1 1 1 1

18 ESP Chambers/ESP No. 1 1 1 1

19 Data Cells/CH No. 1 1 1 1

20 Gas Passages/Cell No. 17 17 17 17

21 Spacing (Plate-to-Plate) in - mm 12 305 12 305

22 Collecting Field 1 ft. - m 14.00 4.27 14.00 4.27

23 Electrodes Field 2 ft. - m 14.00 4.27 14.00 4.27

24 Height ft. - m 22.00 6.71 22.00 6.71

25 Design Coll. Area/Precip. sq. ft. - m2 20,944 1,946 20,944 1,946

26 Details Total ESP El. Length ft. - m 28.00 8.53 28.00 8.53

27 Total Collecting Area sq. ft. - m2 20,944 1,946 20,944 1,946

28 SCA sq. ft./k acfm - m2/m3/s 387.78 76.33 387.78 76.33

29 Calculated Aspect Ratio 1.27 1.27 1.27 1.27

30 Values Face Velocity ft./s - m/s 2.41 0.73 2.41 0.73

31 Treatment Time s 11.63 11.63 11.63 11.63

32 Efficiency Migration Velocity (D/A),cm./s cm./s 5.12 5.13 5.12 5.13

33 Data Modified migration vel., (k=.5) cm./s 20.05 20.05 20.05 20.05

34 Estimated Eff. % 98.00 98.00 98.00 98.00

35 Outlet Fly Ash gr./scf - mg/Nm3, wet36 Conditions Concentration gr./acf - mg/m3 0.0712 162.98 0.0712 162.98

37 Mass Emissions lb/MBtu

38 Stack Exit Stack Dia. ft. - m

39 Opacity Computer Estimated %


34/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONS3.3.3 Model Setup.

The computational mesh has been constructed using the specifica-

tions provided by the customer utilizing the body-fitted coordinates ap-

proach, which allows the creation of the grid of complex geometry. Thestructure of the computational cells (computational mesh) is depicted on

Figures 1 and 2 below. Only one half of the total electrostatic precipitator

ductwork has been actually simulated, this is possible due to symmetry

with respect to the vertical plane passing through the center of the elec-

trostatic precipitator. The computational grid consisted of 81 x65 x 18 or94,770 computational cells. The distinct parts of the computation are de-

picted in Figure 3, which shows the position of the internal details in the

hot gas electrostatic precipitator CFD model.

The flow velocity vectors are depicted in the central vertical section

of the electrostatic precipitator (Figure 5). These were also computed in

the sections (planes) of the precipitator shown on the Figure 4. The re-

sults are represented by Figures 6 through 13. The last two figures (Fig-

ure 12 and Figure 13) show the distribution of the gas flow in the planes

where parameters of this distribution were calculated. These planes are

numbered 7 and 8 on Figure 4.

3.3.4 Results Discussion.

This study represents results of the CFD simulation of the hot gas

precipitator with an additional gas distribution device installed at the preci-

pitator inlet. The latter is of variable aperture or tapered design, for it has

60% open area at the top gradually converging to 40% open area at the

bottom. The purpose of this grid is to redistribute the gas flow in the pre-

cipitator upward to assure lower gas velocities at the bottom in the hopper

area to further improve overall precipitator collection efficiency.

Analyzing velocity vectors on Figures 5 through 13, it could be see

that number of vortices is being formed inside the hot gas electrostatic


35/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSprecipitator. The flow of the gas is entering the system from the above,

and, as a result, has a rather significant downward component present

throughout the electrostatic precipitator. While the horizontal guiding

vanes at the entrance of the electrostatic precipitator redistribute this flow

to a certain degree, they are unable to eliminate this downward flow. Big

portion of the flow is directed downward to the hopper section, this flow

hits the hopper baffle and starts to recirculate. The recirculation leads to

high velocities of the gas flow near the precipitator front with respect to the

flow direction wall of the hopper, these large velocities could be seen on

Figures 5 6 and 9. Even the presence of the variable aperture gas dis-

tribution grid does not eliminate the re-circulation of the gas in the up-

stream part of the hopper. The variable aperture grid does appear to

improve the gas distribution in both, vertical and horizontal planes.

Some recirculation is still present near the precipitator exit behind

the gas distribution grid (see, for example, Figures 5, 7) as it was in the

previous CFD study (Report 103-98 A). Overall, the grid does improve

gas distribution at the inlet portion of the precipitator. However, its impact

diminishes towards the rear (downstream) part of the precipitator.

3.3.5 Summary and Conclusions

The interaction between gas distribution and precipitator physics is

complex, and the conventional wisdom demands that the gas distribution

should be very even, which is mostly based on the precipitator efficiency

formulas. In the first filed, where most coarse particles are found, there is

no sense in trying to raise the heavy dust fraction to the upper region, as it

finally has to be accumulated in the hoppers. This supports as distributionwith velocities above average at the lower part of the field. From the

second filed, it is the finer particles, which are to be caught, suggesting the

gas distribution should be even until the exit from the last field. Particle

reentrainment in the lower part of the last field might be swept out in the

clean gas duct, and it is recommended that the bottom outlet velocities be


36/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSbelow average. Outlet transitions fitted with gas distribution devices make

it possible to adjust the vertical velocity profile, which is essentially impor-

tant in cases where the precipitator is designed with a bottom-type outlet.

The hot gas electrostatic precipitator geometry has been analyzed

numerically with the help of the Computational Fluid Dynamics (CFD)

model. This case includes gas distribution grids at the precipitator inlet

and outlet, and several flow direction baffles inside precipitator and in the

hopper. It was recommended to use a variable aperture or tapered de-

sign gas distribution grid at the precipitator inlet with 60% open area at the

top gradually converging to 40% open area at the bottom. The purpose of

this grid is to redistribute the gas flow in the precipitator upward to assure

lower gas velocities at the bottom in the hopper area to further improve

overall precipitator collection efficiency.

Furthermore, the CFD simulation allows for calculation of the gas

velocity deviation. Hot gas distribution analysis inside the precipitator was

also conducted based on the mean gas velocity according to IGCI stan-

dards.

These analyses produced the following results:

Case A Case B

Loca-tion

Deviation, %RMS,

ft2/sec.

Deviation, %RMS,

ft2/sec.

1.15Vav 1.4Vav 1.15Vav 1.4Vav

Front 29.5 21.4 9.32 27.5 20.3 9.29

Back 18.3 9.7 15.4 9.4

Overall, the grid does improve gas distribution at the inlet portion of

the precipitator. However, its impact diminishes towards the rear (down-

stream) part of the precipitator.

Indeed, application of the variable aperture gas distribution device

appears to resolve some of the problems with the gas flow distribution in-

side of the hot gas precipitator. Although the upward re-direction of the


37/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSgas flow did not completely remove the flow recirculation inside the hop-

per, still, overall positive preponderance is present.

Further improvement could be achieved by addressing area of the

small recirculation between the guide vanes in the inlet nozzle (Figure 6).

Although the hot gas recalculation is substantially reduced as compared to

the Case A, this phenomenon could be still improved further. However,

this refinement should be very carefully monitored from the operational

stand point, for the severity of the dust deposits growth in certain areas

could defeat improvements in the gas flow distribution.


38/77




39/77




40/77




41/77




42/77




43/77




44/77



3.4AIR FLOW MODELING IN THE REGENERATIVE THERMALOXIDIZER

3.4.1 Packed Tower

The three-dimensional flow in a cylindrical packed tower was simu-

lated using a CFD model. The inside diameter and the height of the

packed tower measure 8 and 12 feet respectively. The gas flow rate in the

packed tower was rated at 10,000 scfm (at 70 F). The actual gas tem-

perature at the bottom and the top of the packed tower was assumed 213

F and 1,400 F respectively.

In the CFD model, the packed tower was represented by 8,424(18x26x18) cells. The model included the randomly packed media, the

off-centered inlet duct (nozzle), and the exit ductwork as depicted in the

Figure 1.

A uniform mean temperature of 807 F was assumed throughout

the packed tower in the calculation to account for the temperature effect of

the gas. The randomly packed media was simulated as porous material

with permeability specified according to conditions provided by the Smith

Engineering, i.e. that the overall pressure loss across the 7 (seven) feet

deep media shall be 7 inches of water column (WC) at the rated gas flow.

3.4.2 Results

The computed mass flux vectors are plotted in Figures 2 through 4

for three vertical planes in the tower. These results show that the mass

flux inside of the packed bed (between the two horizontal dashed lines is

fairly uniform. According to numerical values, the variation of the mass

flux in the media is less than 7% of the mean flux.

The mean temperature of 807 F was assumed throughout the cal-

culation, and the exact volumetric gas flow rate cannot be accurately cal-


45/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSculated, the mass flux still is a conservative quantity regardless of the

constant temperature assumptions. The uniform mass flux in Figures 2

through 4 implies uniform velocity profile provided that the temperature is

uniform in a horizontal cross section through the packed bed.

Figure 5 depicts the smoke trails corresponding to the flow paths in

the packed tower. The swirling for pattern at the bottom of the tower is

due the off-centered inlet duct (nozzle) design.

Figure 6 shows the calculated pressure distribution across the

packed bed. The pressure differential across the entire bed (0.243 psi or

6.73 in. WC) is consistent with the draft loss in the actual tower (7 in. WC)

3.4.3 Summary and Conclusions

A CFD model has been developed for he packed bed tower with the

off-centered inlet duct. Preliminary computational results show that with

an overall pressure drop of 7 (seven) inches of WC, the mass flow through

the packed bed is uniform. According to numerical values, the variation of

the mass flux in the media is less than 7% of the mean flax.

3.4.4 Figures

Figure 1. Computational Model Geometry Figure 2. Mass-Flux Vector Plot @k=5


46/77



Figure 3. Mass-Flux Vector plot @k=10 Figure 3. Mass-Flux Vector plot @k=15

Figure 5. Smoke Trails Figure 6. Pressure Distribution in the Packed Tower


47/77



3.5MODEL STUDY FOR IN-DUCT BURNERS

3.5.1 Model Set-up

The CFD model was run in a conventional mode to evaluate thegas patterns, velocity vectors, pressure parameters, combustion products,as well as oxygen and NOx formation characteristics.

3.5.2 Discussion

Figures 1 through 12 depict the results of the CFD modeling in thetwo cross-sections:

1. Cross-Section I - along the center line (C/L) of the orifice and alarger hole.

2. Cross-Section II - along the center of the smaller holes, and

3. Cross-Section III - of the plane perpendicular gas flow at theend of the duct.

Figure 1 represents results of the thermal NOx calculations in theCross-Section I. The units are mass fraction in kg/kg, i.e. thermal NOxfraction in kg in total gas flow in kg. Figure 2 depicts the same valuesalong in the Cross-Section II.

Figure 3 depicts the Static Pressure in Pascal in the Cross-Section I,

and the Figure 4 represents the results of the gas density calculations, inkg/m3 in the same cross-section.

Figures 5 and 6 depict the results of the absolute temperature calcu-lations in K.

Figures 7 through 10 depict the Velocity Vectors in Cross-sections Iand II in m/sec. Figures 9 and 10 represent blow-up views to help identifythe flow patterns through the openings in the strands. Although the gasexit profile looks relatively uniform, it could be seen, that there are areaswith extremely high gas velocities virtually completely by-passing the

combustion zones, thus allowing one to suspect the efficiency of theVOC's destruction on those areas (obviously, making an impact on theoverall VOC destruction efficiency). Compare those with the Figure 11,one can see that the gas flow by-passes the high temperature zones andconverges back with the main portion of the gas in the much cooler zoneon the far left side of the plane (in reality it represents let's say a top por-tion of the duct). Perhaps, extending the strands a little bit more could im-prove the flow pattern and, respectively overall VOC destruction efficiency.


48/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONS3.5.3 Figures


49/77




50/77




51/77




52/77




53/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONS3.6LINDBERG INCINERATOR AND ABSORPTION TOWER

FLOW EVALUATION

3.6.1 Model Setup

The CFD software package was utilized to analyze the flow of theexhaust gas throughout an Absorption Tower system. The Chen-Kim -

model of the turbulent flow was applied for these tasks. This model pro-

vides the time-averaged values of velocities and pressure of the air

throughout the system. It has become one of the most popular

two-equation model of turbulence, though mostly for historical rather than

scientific reasons. The model calculates the effective viscosity associated

with the turbulence based on the solution of two equations for the density

of turbulent kinetic energy k and rate of the dissipation of this energy .

Like other two-equation models, the - model is free from the need to

prescribe the length-scale distribution. Although it does necessitate the

solution of two additional differential equations, these are of the same form

as the momentum equation; Fluent (like other modern computer codes)

solves them easily.

The high-Reynolds number - model becomes invalid near thewalls and thus has to be regarded with the set of wall functions (low-

Reynolds number models are also available but rarely used in engineering

community). The logarithmic wall functions have been used here; these

functions are applicable for the Reynolds numbers based on the length of

the system, which are in the range from approximately 106 to 109. The

analysis of the Absorption Tower shows that this is true for almost all sys-

tem considered. The surfaces of the Absorption Tower are assumed

smooth.

The other boundary conditions applied are fixed mass flow rate at

the entrance to the system and fixed pressure at the exit from the system.

The inlet boundary conditions require the prescription of the two turbulent


54/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSparameters, density of the kinetic energy of turbulence and rate of the

dissipation of this energy . The unknown values can be solved for an as-

sumed intensity of the turbulence at the entrance. This allows us deter-

mine the value of , the value of is than determined based on the inlet li-

near dimensions. Typical inlet turbulence intensity is between 1% and

5%, with the high value of 5% has been chosen for our calculations.

The equation of state of exhaust gas was assumed the one of the

constant density gas. This assumption is valid since the static pressure

drop through the system is of order of 1 inch of water, this value is less

than 1% of the atmospheric pressure, and the expected density change is

about one percent. The inlet mass flow rate of the gas and its tempera-

ture and chemical composition has been provided by the customer. The

kinematic viscosity of the gas at the given conditions has been determined

using the Sutherlands formula and was assumed to be the constant

throughout the system.

Three configurations (Cases A through C) of the Absorption Tower

were considered. The first configuration, Case A (details of this configura-

tion are shown on Figure 2) is the original one, which includes the exten-sion of the inlet tube inside tower. The second configuration (Case B,

Figure 2) represents modification where inlet tube ends at the tower en-

trance, and three direction plates and ring and plate with holes are present

inside the tower. The third configuration (Case 3, Figure 2) was proposed

by the customer. This configuration is similar to the Case B configuration

with the exception that perforated ring is absent and perforated plate oc-

cupies the whole cross-section of the tower. The perforated plates were

modeled as porous material with given pressure loss coefficient, this coef-

ficient was determined based on the results of publication G.B.Schubauer,

W.G.Spangenberg and P.S.Klebanoff, Aerodynamic Characteristics of

Damping Screens, NASA TN 2001, January 1950.


55/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSThe computational mesh has been constructed using the specifica-

tions provided by the customer with utilizing the Fluent body-fitted coordi-

nates approach, which allows the creation of the grid of complex geome-

try. It contains 24x14x110 cells (Figure 1). Due to the symmetry of the

Absorption Tower with respect to the central vertical plane, only half of the

exhaust stack was simulated in order to reduce the computational time.

3.6.2 Results and Discussion

The results of the CFD simulations have been provided in the form

of velocity vectors through different sections of the Absorption Tower (Fig-

ures 3-8 for the Case A, Figures 11-16 for the Case B Configuration and

Figures 19-24 for the Case C Configuration). The computational mesh for

the case considered is also presented (Figure 1).

As it could be seen from the figures that the gas flow inside the Ab-

sorption Tower is very complex one with several recirculation zones

present. For the original configuration, the flow recirculates in vertical di-

rection at the bottom of the column, just above the inlet tube (level of the

bottom injection nozzle) and near the exit from the tower (this recirculation

may not present in real tower due to presence of baffles near the tower

exit which are not present in the simulation). In addition to that, the flow

also recirculates in horizontal direction.

The addition of directional plates, removal of inlet tube projection

inside tower and addition of the perforated ring and plate in the Case B

configuration do alter these recirculation zones, but do not eliminate either

of them. Due to high solidity of the perforated plate and ring big portion of

the gas flow go around them, which increases the recirculation above the

plate.

The best in terms of absence of the flow recirculation among the

cases considered is the configuration proposed by the customer. This


56/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSconfiguration does not eliminate the recirculation near the bottom of the

tower in both horizontal and vertical direction, but does eliminate the recir-

culation in this direction in the injection nozzle area.

3.6.3 Conclusions

The CFD simulation allowed inexpensively determine that among

three configurations considered, the configuration presented in the Case C

delivers the best distribution of the gas flow near the injection nozzles.

3.6.4 Figures


57/77


CFD for Engineering Solutions CHAPTER 4. INDUSTRY APPLICATIONSComputational Grid Computational Details


58/77




59/77

ONTARIOPOWERGENERATIONNANTICOKEGENERATIONSTATION

PRECIPITATORIMPROVEMENTPROJECT

Thomas L. Lumley, OPG Nanticoke GSi

Marc D. Voisine P. Eng., Neill & Gunterii

Dr. Henry V. Krigmont, QEP, Allied Environmental Technologies, Inc.

iii

1 INTRODUCTION Ontario Power Generation (OPG) owns and operates the Nanticoke coal-fired generating

station (Nanticoke GS), which is located on the North shore of Lake Erie in Haldimand County.

The station, originally built in the early 1970s, consist of eight (8) units. The units were origi-

nally designed for 505 MW (gross) at maximum continuous rating while firing high-sulfur bitu-minous coal and currently have an adjusted output of 490 MW (gross) based on the approved

fuel plan of a blend of Eastern Appalachian low sulfur and PRB coals. All units are equipped

with electrostatic precipitator (ESP) to collect fly ash.

During the 1980s OPG was faced with reducing sulfur dioxide emissions at the Nanti-coke GS to comply with the new environmental regulations. After careful review of available

technologies and strategies, the station chose to switch to a coal blend of


60/77

phur trioxide and ammonia. Probe biasing is a techniqu whereby each individual injectionprobe is tailored for the flue gas temperature and gas flows local to that probe. This biasing op-

timizes the injection of both SO3 and NH3 into the flue gas and results in a better resistivity mod-

ification and an improved overall system performance.

2.1PATTERNED OR "TAILORED" FGC INJECTIONThe effectiveness of the SO3 and NH3 flue gas conditioning is greatly dependent on flue

gas temperature and flue gas flow. The current system is setup to treat the worst ash condition

and is dependent on boiler load only. This results in over conditioning the flue gas in certain

areas and under conditioning in others. By individually tailoring the injection rates for tempera-ture and flow differentials, better sulphur trioxide and ammonia coverage and utilization can be

achieved. This results in an overall improvement in the effectiveness of the system and a corres-

ponding improvement in precipitator efficiency. Typically, probe biasing will also result in areduction in SO3 and NH3 consumption due to better utilization.

Ideally, the amount of conditioning gas delivered to any single injection nozzle in the flue

gas stream should be proportioned to the amount of ash passing through that nozzle's treatment

area and adjusted for the flue gas temperature at that point. It is entirely possible to make com-prehensive sets of measurements which determine these values, at least for one set of operating

conditions, and it is also possible to divide the delivery of conditioning gas among the injection

nozzles in accordance with such measurements. In practice, a full scientific treatment of thistype is never done, and is rarely even approximated. The reason, aside from the variability of the

target conditions, is that ESP's have a considerable degree of built-in tolerance for random varia-

tions. Witness to this fact is provided by many existing installations where FGC was neither

needed nor used. It is common in these units to find that a wide spread of temperatures acrossthe air preheater outlets is passed through to the ESP inlet along with non-uniform dust concen-

trations, in spite of which the ESP exhibits excellent operation.

In general, inlet conditions which meet industry criteria for satisfactory ESP operation,

allow treatment of a flue gas conditioning system on the basis of uniform temperature and flowdistribution. Another way to state this is that situations which could require special precautions

for distribution of conditioning gases usually also require modifications to permit satisfactory

ESP operation, and incorporation of the latter will usually eliminate the former.

2.2GAS FLOW DISTRIBUTION Another potential upgrade that was identified early in the project definition phase was to

review and improve the flue gas flow distribution on the inlet and outlet of the existing precipita-tor.

In the early 1990s OPG conducted a flow model study on the existing precipitators. As

a result of this study, a number of flow correction devices were installed in the existing precipita-tors. These flow correction devices were for the most part installed on the internal walkways of

the precipitator and serve to skew the flow in the precipitator. Skewing the flow in the precipita-

tor is a flow adjustment technique popularized in the early 1990s. This flow correction tech-nique when properly applied is intended to improve precipitator performance. Although in-

stalled on the Nanticoke precipitators, the resulting performance improvements were mixed and

subjective. Over the years, some of the devices have fallen off or have been removed with littleor no consequences to the performance of the precipitators.


61/77

As a part of the overall particulate collection improvement program, the project team de-cided to remove the previous skewed flow devices and install flow correction devices to provide

an even gas flow distribution across the inlet and outlet of the precipitators in accordance with

industry standards. This technique is the standard for design of new precipitator installations andis documented in the Institute for Clean Air Companies guideline EP-7. The ICAC EP-7

standard is used throughout the industry as the standard for precipitator modeling and when fol-lowed, results in a uniform velocity profile at the inlet and outlet of the precipitators.

3 EQUIPMENT 3.1BOILERS

Nanticoke GS has total of eight (8) opposed-fired B&W radiant boilers rated at 490-512

MW gross (3,600,00 lb/hr steam at 2,450 psig) each. Superheat/Reheat temperature is 1,000 F.

Each boiler-unit has 61,300 ft2

heating surface and 107,700 ft2

tube transfer surface and isequipped with two Ljungstrom SAHs and one Ljungstrom PAH.

Table 1 presents the recent boiler combustion data based on firing of a blend of up to

80% PRB and up to 20% Eastern Appalachian low sulfur coal

3.2FUEL PROCESSINGAND HANDLINGEach boiler is fired by five (5) B&W 10E Pulverizers with the capacity of:

37 tons Eastern Appalachian low sulfur coal or 53 Tons when fueling 100 % Powder River Basin (PRB) coal.

Individual pulverizer(s) are feeding particular rows of the burners (A through E).

Table 1. Boiler Combustion Data

COMBUSTION DATA (@500 MWGROSS) RANGE TYPICALStack gas flow rate (am3/sec) 805 900 850

Stack gas temperature (C) 126 150 138

ESP inlet PM loading (g/am3) 3.1 5.0 3.9

ESP inlet fly ash LOI (%) 5.0 14.0 9.0

Flue gas conditioning system (in service at all times)

Current SO3 injection rate (ppm) 8

Current NH3 injection rate (ppm) 5

3.2.1 FuelCharacteristicsThe units burn a mixture of PRB and Eastern Appalachian low sulphur bituminous coal.

The fuel specification below is based on the respective 80/20% PRB/Eastern Appalachian low-sulfur coal blend.

3.2.1.1 Coal Analyses Table 2 and Table 3 present coal Proximate and Mineral analyses.


62/77

Table 2. Coal Proximate Analysis (as received)

ITEM RANGE TYPICAL

Moisture (%) 14 To 25 23

Volatile matter (%) 30.5 To 34 31

Fixed carbon (%) 39 To 46 40

Ash (%) 5.0 To 8.0 6.0Sulfur (%) 0.3 To 1.0 0.4

HHV (Btu/lb) 9,300 To 11,000 9,450

Table 3. Coal Ultimate Analysis (as received)

ITEM RANGE TYPICAL

Carbon (%) 54 To 64 55

Hydrogen (%) 3.6 To 4.3 3.7

Oxygen (%) 8.5 To 12.0 11.5

Nitrogen (%) 0.7 To 1.2 0.9

Sulfur (%) 0.3 To 1.0 0.4

Ash (%) 5.0 To 8.0 6.0

Moisture (%) 14 To 25 23

3.2.1.2 Ash Mineral Analysis Table 4. Fly Ash Mineral Analyses

ITEM RANGE TYPICAL

Li2O (%)

P2O5 (%) 0.7 To 1.8 1.5

SiO2 (%) 39 To 49 42

Fe2O3 (%) 5 To 11 6.5

Al2O3 (%) 17 To 24 18TiO2 (%) 0.9 To 1.5 1.1

CaO (%) 9 To 18 15

MgO (%) 2.0 To 5.0 4.0

SO3 (%)

K2O (%) 0.5 To 1.3 0.7

Na2O (%) 0.5 To 1.4 1.0

3.3ELECTROSTATIC PRECIPITATORS Each boiler-unit is equipped with a Joy/Western precipitator (Table 5). Each precipitator

has two (2) chambers and three mechanical fields in the direction of the gas flow. The collectingplates are nominally twelve (12) feet long and 30 (thirty) feet tall. Each precipitator is poweredby twelve (12) 1,600 mA sets. All T/R sets are equipped with Belco/Merlin AVCs. There are

80 gas passages nine (9) inch wide in each chamber for a total of 160 gas passages across the gas

flow.

Original design SCA was 225 ft2/1000 acfm at a gas flow of 1,538,000 acfm and 246 F,

however, currently the actual gas flow is typically around 1,800,000 acfm and 280 F, which re-

sults in the SCA of about 192 ft2/1000 acfm.


63/77

Table 5. Precipitator Design Data

ITEM UNITSVALUE

DESIGN ACTUAL

Design Gas Flow Rate acfm 1,538,000 1,800,000

Gas Temperature at the ESP Inlet F 246 280Average Gas Velocity ft/s 7.12 8.4

Specific Collection Area (SCA) ft2/1,000 acfm 225 191

Inlet Loading gr/scf 4.878 2.7

Efficiency % 99.5 98.5-99.05 (w/FGC)

Treatment Time s 5.05 4.3

Current Density mA/ft2

(plate) 0.0556 0.0556

3.4FLUE GAS CONDITIONING 3.4.1 FGCSystems3.4.1.1 SO3 Flue Gas Conditioning System

The original SO3 FGC systems were manufactured by Wahlco, Inc. Each FGC system

injects SO3 independently of the ammonia, based on the Boiler Load signal represented by a coal

flow rate. The design sulfur flow was rated 074 kg/hr as 0-100% on the Ratio Station (control-ler). Maximum SO3 flow was 24 ppmV based on flue gas flow of 2,760,870 Nm

3/hr @127 C

and the Ratio Station signal of 100% on use of in. diameter plunger pump with the pump

stroke set to 1 and inches. The built-in interlock provides the FGC system shut-off at the fuelflow rate below 20%.

Recently, in the process of the upgrading the old piston-type molten sulfur metering

pumps, the latter were replaced with the diaphragm-type LEWA pumps. The deliverable sulfur

flow was also reduced by 50%, thus reducing the maximum SO3 injection rate down to 12ppmV.

3.4.1.2 NH3 Flue Gas ConditioningSystems The NH3 FGC systems were also manufactured by Wahlco, Inc. Each system was de-

signed to inject NH3 independently of the SO3 injection, based on the Boiler Load signalrepresented by a coal flow rate. The design ammonia flow was rated 0 28 kg/hr as 0-100% on

the Ratio Station (controller). Maximum NH3 flow is 20 ppmV based on flue gas flow of

2,760,870 Nm3/hr @127 C and the Ratio Station signal of 100%.

3.4.2 OriginalInjectionProbes3.4.2.1

SO3 Injection Probes

The SO3 injection probes are installed downstream of the SAHs in both pant legs of theexhaust gas flow. Initially it was thought to install the SO3 injection probes in the PAH duct as

well, however, due to the lower gas exit temperatures in this duct, it was decided not to injectSO3 for fear of acid attack in this duct. During initial operation the SO3 injection probes operat-

ed very successfully. However, over time, the probes were prone to pluggage. Consequently,

the probes were modified. The latter included shortening the probes and adding a single largernozzle pointing downwards to prevent dust build-up and subsequent plugging. This new SO3

probe design is the standard at the station.


64/77

3.4.2.2 NH3 Injection Probes The NH3 injection probes were installed downstream of the secondary air heaters (SAH)

and upstream of the SO3 injection probes. There were no NH3 injection probes in the primary air

(PAH) duct.

Figure 1. Unit 2 SO3 Distribution Down-

stream of the Injection Probes

Figure 2. Unit 2 SO3 Distribution in a Hori-

zontal Plane

Figure 3. Unit 5 SO3 Distribution Down-

stream of the Injection Probes

Figure 4.Unit 5 SO3 Distribution in a Hori-

zontal Plane

4 WORKPERFORMED 4.1FLUE GAS CONDITIONING UPGRADES4.1.1 Design

To determine the number of new SO3 and NH3 probes required in the PAH duct and de-sign of the probe biasing, a computational fluid dynamics (CFD) modeling was performed. The

CFD included modeling the flue gas flow and temperatures distribution from the air pre-heaters

outlet, to the precipitators inlet. From the CFD model, the proper number of SO3 and NH3probes were determined as well as the optimum injection rates for all the probes to ensure a uni-

form distribution of SO3 and NH3 across the inlet of the ESP.


65/77

4.1.2 SO3SystemBased on the CFD model it was determined that four (4) new SO3 probes would be re-

quired for Unit 2 and two (2) SO3 probes would be required for Unit 5 (Figure 1 and

Figure 2). The latter was done in anticipation of a slightly different temperature distribution pro-file somewhat similar to the one on Unit 7 due to the new deeper baskets in the Unit 5

PAH. However, there was a provision made to install an additional two (2) probes on the PAHduct of Unit 5 if required. As shown on Figure 3 and Figure 4, although there is a provisionf

statement of qualifications-cfd

Documents